R Notebook

Bonus visualization or comparisons

library(dplyr)

G3;
Attaching package: ‘dplyr’

gG3;The following objects are masked from ‘package:stats’:

    filter, lag

gG3;The following objects are masked from ‘package:base’:

    intersect, setdiff, setequal, union

g

Visualize feature counts and variability

log10 = 0 is equale to a variance of 1 log10 = -2 is 0.01 log10 = -4 is 0.0001

ft <- gene_vars[gene_vars > 0]
quant <- quantile(gene_vars, 0.95)
hist(ft[ft < quant], breaks = 100, col = "steelblue",
     main = "Gene Variance (95% trimmed)", xlab = "Variance")

quant <- quantile(gene_vars, 0.90)
hist(ft[ft < quant], breaks = 100, col = "steelblue",
     main = "Gene Variance (90% trimmed)", xlab = "Variance")

quant <- quantile(gene_vars, 0.75)
hist(ft[ft < quant], breaks = 100, col = "steelblue",
     main = "Gene Variance (75% trimmed)", xlab = "Variance")

NA
NA
NA
NA
NA

ggplot(df, aes(x = mean, y = variance)) +
  geom_point(alpha = 0.3, color = "steelblue") +
  scale_x_log10() +
  scale_y_log10() +
  labs(title = "Gene Dispersion", x = "Mean Expression", y = "Variance") +
  theme_minimal()


ggplot(df, aes(x = mean, y = variance)) +
  geom_point(alpha = 0.3, color = "steelblue") +
  labs(title = "Gene Dispersion", x = "Mean Expression", y = "Variance") +
  theme_minimal()



ggplot(df, aes(x = mean, y = variance)) +
  geom_point(alpha = 0.3, color = "steelblue") +
  scale_y_log10() +
  labs(title = "Gene Dispersion", x = "Mean Expression", y = "Variance") +
  theme_minimal()

ggplot(df_ft, aes(x = mean, y = variance)) +
  geom_point(alpha = 0.3, color = "steelblue") +
  scale_x_log10() +
  scale_y_log10() +
  labs(title = "Gene Dispersion", x = "Mean Expression", y = "Variance") +
  theme_minimal()

ggplot(df_ft, aes(x = mean, y = variance)) +
  geom_point(alpha = 0.3, color = "steelblue") +
  labs(title = "Gene Dispersion", x = "Mean Expression", y = "Variance") +
  theme_minimal()

ggplot(df_ft, aes(x = mean, y = variance)) +
  geom_point(alpha = 0.3, color = "steelblue") +
  scale_y_log10() +
  labs(title = "Gene Dispersion", x = "Mean Expression", y = "Variance") +
  theme_minimal()

NA
NA

# cumulative variance plot
sorted_vars <- sort(gene_vars, decreasing = TRUE)
cumvar <- cumsum(sorted_vars) / sum(sorted_vars)

plot(cumvar, type = "l", lwd = 2, col = "darkgreen",
     xlab = "Top N Genes", ylab = "Cumulative Variance Explained",
     main = "Cumulative Variance Plot")

How “vst” variance colculaton works

Mean variance plot method

ggplot(df_ft, aes(x = mean, y = dispersion)) +
  geom_point(alpha = 0.3, color = "steelblue") +
  scale_x_log10() +
  scale_y_log10() +
  labs(x = "Log10 Mean Expression", y = "Log10 Dispersion") +
  theme_minimal()


# Set your MVP thresholds
mean_cutoff <- c(0.1, 8)
dispersion_cutoff <- c(1, 50000)  # Inf will not be plotted as a line

# Make the plot
ggplot(df_ft, aes(x = mean, y = dispersion)) +
  geom_point(alpha = 0.3, color = "steelblue") +
  scale_x_log10() +
  scale_y_log10() +
  # Add vertical lines for mean cutoff
  geom_vline(xintercept = log10(mean_cutoff[1]), linetype = "dashed", color = "darkred") +
    geom_vline(xintercept = log10(mean_cutoff[2]), linetype = "dashed", color = "darkred") +
  # Add horizontal lines for dispersion cutoff (excluding Inf)
  geom_hline(yintercept = log10(dispersion_cutoff[1]), linetype = "dashed", color = "darkgreen") +
  labs(x = "Log10 Mean Expression", y = "Log10 Dispersion") +
  theme_minimal()




ggplot(df_ft, aes(x = mean, y = dispersion)) +
  geom_point(alpha = 0.3, color = "steelblue") +

  # Add vertical lines for mean cutoff
  geom_vline(xintercept = mean_cutoff[1], linetype = "dashed", color = "darkred") +
    geom_vline(xintercept = mean_cutoff[2], linetype = "dashed", color = "darkred") +
  # Add horizontal lines for dispersion cutoff (excluding Inf)
  geom_hline(yintercept = dispersion_cutoff[1], linetype = "dashed", color = "darkgreen") +
  labs(x = "Mean Expression", y = "Dispersion") +
  theme_minimal()

NA
NA

Plot dispersion

Explore Variable Feature Selection Overlap

seu <- FindVariableFeatures(seu, selection.method = "dispersion",
  loess.span = 0.3,
  clip.max = "auto",
  num.bin = 20, 
   nfeatures = 2000,
  binning.method = "equal_width",
  mean.cutoff = c(0.1, 8),
  dispersion.cutoff = c(1, Inf),
  verbose = TRUE)

G3;Finding variable features for layer data
gCalculating gene means
0%   10   20   30   40   50   60   70   80   90   100%
[----|----|----|----|----|----|----|----|----|----|
**************************************************|
Calculating gene variance to mean ratios
0%   10   20   30   40   50   60   70   80   90   100%
[----|----|----|----|----|----|----|----|----|----|
**************************************************|

Var_ft_dis <- VariableFeatures(seu)
seu@misc$Var_ft_dis <- Var_ft_dis


VariableFeatures(seu) <- seu@misc$Var_ft_dis
# Identify the 10 most highly variable genes
top10dis <- head(VariableFeatures(seu), 10)

# plot variable features with and without labels
plot3 <- VariableFeaturePlot(seu, assay = "RNA", selection.method = "dispersion",top10dis, repel = TRUE)

G1;H1;Errorh in VariableFeaturePlot(seu, assay = "RNA", selection.method = "dispersion",  : 
  unused argument (repel = TRUE)
Error during wrapup: not that many frames on the stack
Error: no more error handlers available (recursive errors?); invoking 'abort' restart
g

LabelPoints(plot = plot2, points = top10dis, repel = TRUE)

G3;When using repel, set xnudge and ynudge to 0 for optimal results
gG2;H2;Warningh: Removed 2 rows containing missing values or values outside the scale range (`geom_point()`).g

make an upset plot comparing the 3 FindVariableFeatures options

upset(fromList(list(vst = Var_ft_vst, mvp = Var_ft_mvp, disp = Var_ft_dis)))

NA
NA

Make an upset plot for the variable features for each sample and the merged object

# example
upset(fromList(list(C1 = Var_C1, C2 = Var_C2, C3 = Var_C3, C4 = Var_C4, Combined = Var_obj)))